Target Reacquisition in Aerial Videos
In this project, we attempt to solve the problem of persistent tracking in aerial imagery by using novel ideas of Motion and Appearance Contexts. Motion context is a collection of trajectories belonging to objects in the given environment, which are representative of the motion of the occluded or unobserved object. The appearance context of an occluded object consists of appearance information of all other occluded or unobservable objects in the environment. This appearance context is incorporated by learning a distribution of interclass variation for each target-unobservable object pair.
Saad Ali, Vladimir Reilly, and Mubarak Shah, Motion and Appearance Contexts for Tracking and Re-Acquiring Targets in Aerial Video, IEEE Conference on Computer Vision and Pattern Recognition, Minneapolis, USA 2007.